wonderful strife: systematics, stem groups, and the...

q 2005 The Paleontological Society. All rights reserved. 0094-8373/05/3102-0007/$1.00

Paleobiology, 31(2), 2005, pp. 94–112

Wonderful strife: systematics, stem groups, and the phylogeneticsignal of the Cambrian radiation

Derek E. G. Briggs and Richard A. Fortey

Abstract.—Gould’s Wonderful Life (1989) was a landmark in the investigation of the Cambrian ra-diation. Gould argued that a number of experimental body plans (‘‘problematica’’) had evolvedonly to become extinct, and that the Cambrian was a time of special fecundity in animal design.He focused attention on the meaning and significance of morphological disparity versus diversity,and provoked attempts to quantify disparity as an evolutionary metric. He used the Burgess Shaleas a springboard to emphasize the important role of contingency in evolution, an idea that he re-iterated for the next 13 years. These ideas set the agenda for much subsequent research. Since 1989cladistic analyses have accommodated most of the problematic Cambrian taxa as stem groups ofliving taxa. Morphological disparity has been shown to be similar in Cambrian times as now. Kon-servat-Lagerstatten other than the Burgess Shale have yielded important new discoveries, partic-ularly of arthropods and chordates, which have extended the range of recognized major clades stillfurther back in time. The objective definition of a phylum remains controversial and may be im-possible: it can be defined in terms of crown or total group, but the former reveals little about theCambrian radiation. Divergence times of the major groups remain to be resolved, although mo-lecular and fossil dates are coming closer. Although ‘‘superphyla’’ may have diverged deep in theProterozoic, ‘‘explosive’’ evolution of these clades near the base of the Cambrian remains a pos-sibility. The fossil record remains a critical source of data on the early evolution of multicellularorganisms.

Derek E. G. Briggs. Department of Geology and Geophysics, Yale University, Post Office Box 208109,New Haven, Connecticut 06520-8109. E-mail: [email protected]

Richard A. Fortey. Department of Palaeontology, The Natural History Museum, Cromwell Road, LondonSW7 5BD, United Kingdom. E-mail: [email protected]

Accepted: 24 April 2004

Introduction

Of the popular books written by S. J. Gould,none enjoyed more global success than Won-derful Life (1989). The book, at heart, was a de-tailed account of the investigation of the Mid-dle Cambrian ‘‘soft-bodied’’ fossils from theBurgess Shale of British Columbia discoveredby Charles Doolittle Walcott 80 years earlier. Itcelebrated the kind of patient morphologicalwork carried out by Harry Whittington andhis colleagues over several decades, and formany paleontologists provided a high-profilevindication of what might be termed ‘‘classi-cal’’ research. Thanks to Wonderful Life, suchesoteric animals as Hallucigenia and Anomalo-caris became almost as familiar to the scientif-ically literate as Brontosaurus. Gould also pro-posed several major new ideas, some derivedfrom the work of the students of the Cambrianfossils, others entirely his own. These ideasstimulated criticism and research programs inequal measure. Fifteen years later, it is timely

to examine how Gould’s ideas on the Cambri-an have fared in the light of subsequentthought and discoveries. Whatever the out-come, few would question that Wonderful Lifeserved as an inspiration at best, a goad atworst. It was a catalyst for the examination ofissues that are still under discussion today.

The main focus of Wonderful Life was themeaning of the Cambrian evolutionary ‘‘ex-plosion.’’ Gould apparently had no doubt thatthe appearance of major groups of animals atthe base of the Cambrian was a product ofrapid evolution of novel clades, rather than,for example, a change in their fossilization po-tential due to the acquisition of skeletons andan increase in body size. The notion of an un-rivaled period of morphological experimen-tation in the Cambrian led to a controversialcorollary: that the range of designs—whatGould termed ‘‘disparity’’—at some funda-mental level was actually greater in the Cam-brian than it was subsequently.

Although early claims about the numbers of

95THE CAMBRIAN RADIATION

Cambrian body plans are not taken seriouslytoday, the phylogenetic ‘‘explosiveness’’ orotherwise at the base of the Cambrian has re-mained an issue ever since, with new fossildiscoveries and new kinds of evidence, nota-bly molecular, being brought to bear on thetopic. What range of form evolved in the Cam-brian, and how quickly? How should we clas-sify fossils that appear to fit on the stem-group leading to a ‘‘modern’’ clade? What, af-ter all, is a phylum, and how do we recognizeone? How long was what Cooper and Fortey(1998) termed the ‘‘phylogenetic fuse’’ prior tothe ‘‘explosion’’? If the range of designs wasso wide in the Cambrian then the chance ex-tinction of one animal rather than anothermight have served to reset the subsequent di-rection of evolution. To what extent is theshape of life contingent upon the chance ex-tirpation of what might have proved to be an-other great group in statu nascendi? We mightexpect answers to some of these questions bynow, not least because many new fossil faunashave been discovered since Wonderful Life waswritten. In this paper we examine the extentto which Gould’s views have stood the test ofnew evidence, and discuss how subsequent re-search has raised new questions since his as-sessment of the meaning of the Burgess Shale.

Classification of the Burgess Shale Animals

Gould championed the significance of fos-sils in understanding the early history of an-imal groups and their interrelationships, aview that contrasted with the stance taken bymore extreme ‘‘pattern cladists’’ who arguedthat fossils can provide little or no useful in-formation about the phylogeny of modern or-ganisms (Patterson 1981). In Wonderful Life(1989) Gould emphasized the importance offossils, and particularly Cambrian fossils, inproviding evidence of morphologies that havebeen lost as a result of extinction. The initialemphasis of interpretations of the BurgessShale fossils was on their peculiarities. Therewere indeed aspects of the design of some ofthe animals that were unique. These fossilsdid not—still do not—fit readily into catego-ries based solely on the living fauna. At thetime Wonderful Life was published, a currentperception was that the peculiarities that such

animals exhibited was an indication that theymust have originated separately from differ-ent putative soft-bodied Precambrian ances-tors. This evolved into a view that at leastsome of the ‘‘weird wonders,’’ such as Hallu-cigenia and Opabinia, were preserved frag-ments of ‘‘new phyla’’—experiments in designthat failed to prosper, and were weeded out bysubsequent extinctions. This view of evolutioncame under scrutiny almost from the outset(Fortey 1989; Briggs and Fortey 1989), andmuch progress has been made subsequently infitting even some of the most bizarre animalsinto a reasonable tree of descent, without in-voking polyphyly. Progress toward under-standing the phylogenetic relationships andclassification of Cambrian soft-bodied ani-mals has come from various sources.

Critical restudy of some of the animalsshowed that they were not as peculiar as firstclaimed. The cause celebre in this regard wasHallucigenia, a definitively peculiar animal‘‘with an anatomy to match its name’’ (Gould1989: p. 14) which, when further prepared(Ramskold 1992), showed paired series of legsthat demonstrated its nature as some kind oflobopod. In the process, the original recon-struction was turned upside down, and its‘‘unknown phylum’’ status relinquished. Sim-ilar attempts were made to rehabilitate theenigmatic Opabinia (Budd 1996) as a stem-group arthropod, by reinterpretation of its ap-pendages and the identification of leglikestructures on its ventral surface, although theevidence for these remains equivocal.

New discoveries served to link formerlyproblematic taxa more closely with recog-nized clades, revealing the possibility of theirrational classification. This was particularlythe case when new taxa from the Chengjiangfauna of China and the Sirius Passet fauna ofGreenland added to the picture of Cambriandiversification. It had been appreciated sinceConway Morris’s (1977b) study of priapulidsfrom the Burgess Shale that some groups ofminor significance today were of considerableimportance and diversity in the Cambrian.Following his 1993 description of Kerygmachelafrom Greenland (Budd 1993), Budd arguedthat the lobopods were also comparatively di-verse in these early faunas and made a case for

96 DEREK E. G. BRIGGS AND RICHARD A. FORTEY

their seminal role in the evolution of arthro-pods (Budd 2002). Whether or not his ideasprove to be correct in detail, they serve to linka variety of ‘‘weird’’ fossils with more familiarmorphologies. Some other enigmatic Cambri-an fossils also turned out to be lobopods—forexample, the plates known as Microdictyon arethe ‘‘body armor’’ of one of these creatures(Chen et al. 1989; Ramskold and Chen 1998).Other fossils that first appeared to be uniqueto the Burgess Shale proved to have close rel-atives from other Cambrian Konservat-Lag-erstatten, thus turning an oddity into a clade.The giant animal Anomalocaris excited muchspeculation as to whether it was, or was not,an arthropod. It had an arthropod-like pair ofcephalic grasping limbs, but its trunk append-ages appeared unique. Gould, followingWhittington and Briggs (1985), consideredthat it represents a separate phylum. The dis-covery of anomalocaridids from China (Houet al. 1995), and additional forms from BritishColumbia that remain to be described (Collins1996) increased the disparity of the group, andfor all their peculiarities, brought the anom-alocaridids within the compass of the discus-sion of arthropod origins (definitive evidenceof segmented trunk limbs in anomalocarididsremains to be published—Parapeytoia may bea ‘‘great-appendage’’ arthropod). At the sametime, these new animals suggested that theanomalocaridids may have constituted aclade, Dinocarida (Collins 1996). Whether Hal-kieria (Conway Morris and Peel 1995) fromGreenland is a fossil that links molluscs andbrachiopods is still under debate, but it can beadded to the inventory of Cambrian animalsthat provide evidence of character combina-tions that link high-level taxa rather than sug-gesting ‘‘separate origins’’ from some un-known ancestor.

Gould did not embrace the cladistic method(Levinton 2001), relying on a more traditionalapproach to phylogeny, which tended to em-phasize differences between taxa rather thanthe similarities that indicate relationship.Many of the Burgess Shale animals were suf-ficiently well preserved to be incorporatedinto cladistic analyses of relationship. Withtheir complex characters, arthropods were themost suitable candidate for combining data

from fossils and Recent animals together in asingle analysis. Wills et al. (1994) demonstrat-ed that the great majority of Cambrian taxacan be accommodated with their moderncounterparts in a plausible hypothesis of re-lationships—by building a phylogeny fromthe bottom up rather than trying to place fos-sils into higher taxa defined exclusively on thebasis of living organisms. This approach shift-ed the emphasis from autapomorphy—themuch-bruited ‘‘weirdness’’—to synapomor-phy, the characters shared between one ani-mal and another. The logic in ranking of taxademands that a classification is based on ananalysis of this kind, in contrast to Hou andBergstrom’s (1997) erection of 16 new highertaxa (as well as six new families) when de-scribing the Chengjiang arthropods, withoutrigorous character analysis. The first cladisticanalysis that attempted to order the BurgessShale arthropods by using parsimony (Briggsand Fortey 1989) has been followed by manyanalyses using progressively more sophisti-cated character sets (e.g., Wills et al. 1998) thatcombine fossil and living taxa. At the sametime, phylogeny of the arthropods as awhole—including sister taxa such as Onycho-phora and Tardigrada—has been being tack-led from the molecular systematics database(Wheeler 1998). Increased levels of resolutionhave been obtained by analyzing molecularand morphological evidence in a single data-base (Giribet et al. 2001). Fossil-based cladis-tics, and the ‘‘top down’’ approach of usingmolecular sequences from living animals,constitute a double assault on the arthropodtree. Even so, it would be incorrect to claimthat even the major features of the tree are un-controversial; for example, different versionsof the deep branches in arthropod phylogenywere published in adjacent papers in Nature(Hwang et al. 2001; Giribet et al. 2001), andpaleontologists continue to disagree about thehomologies to be recognized in coding mor-phological characters.

Phylogenetic analyses of the Burgess Shalearthropods based on parsimony (Briggs andFortey 1989; Briggs et al. 1992a; Wills et al.1994, 1998) were based on all available mor-phological characters. Whether the characterswere unweighted, or weighted on the basis of


successive approximations, the gross topolo-gy was retained. However, consistency indiceswere relatively low and the positions of someindividual taxa fluctuated, perhaps as a nat-ural consequence of the character combina-tions found in the early representatives ofgroups. We do not understand enough yetabout the homology and polarity of charactersto determine the phylogeny of the Cambrianforms in detail. Developmental studies ofmodern arthropods offer the best hope for de-termining the significance of characters andhow to use them.

Regardless of ongoing contention, a num-ber of generalizations revealed by the cladisticapproaches are relevant to the scenario out-lined in Wonderful Life. Comparatively few fos-sils have a morphology that prevents their be-ing incorporated into a tree with living forms.Most of the arthropod genera can be consid-ered as stem crustaceans or stem arachno-morphs, and there is good consensus on theplacing of most of the important taxa. The Tri-lobita, for example, are securely placed withina larger arachnomorph clade. Dinocarida forma sister group of Euarthropoda. Onychophoralie outboard of the euarthropods. Since thework of Averof and Akam (1995) Insecta andCrustacea are increasingly regarded as allied,and the former have neither any Cambrian re-cord nor plausible sister taxon of that age. It isnot necessary to invoke phylum-level taxa toaccommodate the Cambrian fossil material.However, the autapomorphies exhibited bysome of the early arthropod stem groups areopen to different interpretations: some work-ers consider them ‘‘worth’’ a high-level taxon,whereas cladistic analyses show that many ofthem constitute plesions on the tree leading tothe crown group.

The Question of Disparity

As a response to what might be termed cla-distic reductionism, Gould (1991) emphasizedthe quantification of morphospace as a mea-sure of ‘‘disparity’’; i.e., he considered themorphological reach of autapomorphies to beas significant as the collective compass of syn-apomorphies. In a universe of possible de-signs—without prime regard to phylogeneticorigin—was he correct in asserting that the

Cambrian forms occupied a larger volume ofmorphospace than living animals? Does theevidence of the Burgess Shale animals invertthe conventional cone of increasing diversity?This turned out to be a hard question to ad-dress, not least because of the difficulties ofquantifying design (Briggs et al. 1992a). Of theCambrian faunas, the arthropods are perhapsthe most obvious group to use in tackling thisquestion, because they exhibit striking differ-ences in appendages and tagmatization. Willset al. (1994) used principal component analy-sis to determine the distribution of a range ofliving and Cambrian arthropods in morphos-pace, supplemented by a variety of measuresof disparity, and concluded that there was nodifference in morphospace compass betweenCambrian and Recent faunas. The cone of in-creasing diversity was not inverted (sensuGould 1989: p. 46) but might be representedrather as a ‘‘tube’’ with a diameter that re-mained roughly constant following its estab-lishment in the early Cambrian (see Wills andFortey 2000). There was certainly no evidencefor greater disparity of design in the Cambri-an, one of the main conclusions of WonderfulLife. The analysis of Wills et al. (1994) was vul-nerable to the criticism that the Recent ani-mals had been selected for their taxonomic‘‘spread’’ rather than randomly sampled—such as was supposed to have happened in thefossilization process (Briggs et al. 1992b; Footeand Gould 1992; Lee 1992). Lofgren et al.(2003) showed that disparity among Carbon-iferous arthropods was approximately 90% ofthat in both the Cambrian and Recent, butthey emphasized that the important differenceis not the amount of morphospace but the par-ticular regions occupied through time. A sec-ond study by Wills (1998) on the priapulidworms was able to sample more comprehen-sively the Recent fauna of this species-poorgroup, and found that the Cambrian taxa oc-cupied less morphospace than the Recent. Inthis case, there was also a shift in the mor-phospace regions occupied between the Cam-brian and Recent. There is no study that re-ports to the contrary, and enthusiasm for‘‘greater disparity’’ in the Cambrian appearsto be on the wane. Discoveries of younger, un-usual morphologies in the Silurian Konservat-


Lagerstatte of Herefordshire, England (e.g.,Sutton et al. 2001) and the Devonian Huns-ruck Slate of Germany (Briggs and Bartels2001), which had already yielded Mimetaster,part of a clade with Marrella, also serve to di-minish the apparent distinctiveness of theCambrian faunas. The lack of what Gouldwould have called weird wonders in the Or-dovician and Silurian may be at least partly amatter of the scarcity of exceptional preser-vation (Allison and Briggs 1993; Briggs et al.1996). This more prosaic view of the Cambrianradiation is now widely accepted but, even ifdisparity were not greater during the Cam-brian, by that time the radiation had filled asmuch morphospace as is occupied by all thearthropods today. The rapid evolution of formstill remains to be explained, and it is only theamount of evolution implied by Gould’s viewthat has diminished.

What, If Anything, Is a Phylum? TheCrown Group–Stem Group Debate

Gould’s (1989) thesis on the Cambrian ra-diation was based around the concept of phylaor body plans. The focus on phyla has shiftedfrom considering them as a measure of mor-phological separation to attempting to under-stand their relationships, and much of thishinges on the use of molecular data. However,the problem of conceptualizing phyla contin-ues to permeate debates about the nature ofthe Cambrian explosion. Budd and Jensen(2000) reassessed the fossil record of the bi-laterian phyla in an attempt to clear up ‘‘con-fusion in the definition of a phylum’’ (p. 253).In simple terms their view was that the earliestmembers of a lineage leading to the modernmembers of any phylum cannot, by definition,have acquired the diagnostic body plan, andtherefore cannot be identified, on the basis ofmorphology, as belonging to that particularphylum. So at what point does a lineage meritthe term phylum?

Budd and Jensen (2000; see also Budd 2003)resorted to a crown-versus stem-group defi-nition (Fig. 1A), a method provided much ear-lier (Jefferies 1979) to accommodate both fossiland Recent taxa in cladograms. The crowngroup within a clade consists of the last com-mon ancestor of all living forms in the phylum

and all of its descendants. The stem group isthe remainder of the clade, i.e., a series of ex-tinct organisms lying ‘‘below’’ the crowngroup on the cladogram. Budd and Jensen ar-gued that the origin of the crown group is theonly point in the cladogram that can be objec-tively defined, so they equated the appearanceof the phylum with the origin of the crowngroup.

Defining the crown group as the phylum isconvenient mechanistically, as it is straightfor-ward to apply and avoids the difficulty of howto separate members of sister phyla near theirdivergence from a common ancestor. It focus-es on that part of the tree of life that includesextant animals and for which, therefore, mo-lecular evidence is available. Restricting theconcept of a phylum to the crown group as ad-vocated by Budd and Jensen (2000) also has anumber of drawbacks, however, that reduceits utility as a measure of the timing and evo-lution of character complexes (Wills and For-tey 2000). There are, by definition, no extinctphyla—only clades with living representa-tives merit phylum status. Curiously, then,some fossil taxa do not belong to a phylum.Membership of the crown group varies withdifferent hypotheses of relationship (althoughit is not a phylum, the crown clade Mammaliaprovides a good example [Benton 2000]). Phy-la are defined on the basis of an arbitrarily se-lected time line (the Recent), the autapomor-phies present in the living forms determinehow many and which characters define thephylum, and fossils that do not share these au-tapomorphies are not members. The inclusionof fossil taxa within the crown group usuallydepends on the chance survival of a few prim-itive organisms today, and this in turn inevi-tably determines how early the phylum orig-inates. The survival of modern horseshoecrabs, for example, pulls the extinct eurypter-ids into the crown clade of chelicerates (Seldenand Dunlop 1998). On this basis a phylum isdefined by the possession of an arbitrarynumber of autapomorphies, and an organismwith one fewer (n 21 rather than n), lying im-mediately below the crown group, is disqual-ified (Fig. 1A). Were the horseshoe crabs to goextinct, the chelicerate crown group would bedefined by scorpions 1 sister taxa; eurypter-


FIGURE 1. Diagrammatic illustration of two definitions of phyla. A, The crown-clade definition advocated by Buddand Jensen (2000). Phylum X contains three taxa, E to G, of which one is extinct. Extinct taxa A to D belong to thestem group, but are not part of the phylum. The phylum is defined by five synapomorphies; extinct taxon D, whichhas just one fewer, is disqualified from membership. Similar considerations apply to Phylum Y. B, The total-cladedefinition, preferred here. Phylum X contains seven taxa, living and extinct, and is defined by one synapomorphy.Similar considerations apply to Phylum Y. Taxa A and H may be difficult to assign, as their proximity to the originis reflected in similarities of primitive morphology, but this difficulty does not bear on the reality of their positionto one side of the node or the other. Note that the taxa in the diagram could represent any level below phylum, and‘‘synapomorphy’’ implies a suite of characters that define a node. (Based on Jefferies 1979; Budd and Jensen 2000;Wills and Fortey 2000.)

ids and horseshoe crabs would be excluded,despite the many characters they share withthe scorpions. There seems something very ar-bitrary about defining a major group on a

whim of history. Budd and Jensen’s definitionimplies that neither the Phylum Arthropodanor the Mollusca are represented until theLate Cambrian, the Annelida until the Ordo-


vician, or the Priapulida until the Carbonif-erous, although Chordata and Brachiopodastill originate in the Early Cambrian. But fewwill agree that this helps us to track the evo-lution of major groups; most still adhere toGould’s (2002: p. 1155) view that ‘‘all major bi-laterian phyla with conspicuously fossilizablehard parts make their first appearance in thefossil record within a remarkably short inter-val . . . of the so-called Cambrian explosion.’’

Budd and Jensen (2000) reduced the signif-icance of the Cambrian ‘‘explosion’’ not bymaking more time available for evolution inthe Precambrian (the cryptic fossil record im-plied by some molecular clock estimates), norby diminishing the amount of evolution re-quired (by quantifying and comparing mor-phological separation among sample taxa),but by arguing that much of the morphologi-cal evolution required to give rise to the mod-ern phyla actually took place later than is nor-mally acknowledged, during the Phanerozoic.On the other hand, the more new Cambrianarthropods that are discovered, the more(morphologic) evolution seems to have hap-pened already by the early Cambrian. Al-though we endorse the need to interpret thehistory of body plans in the light of well con-strained phylogenies, Budd and Jensen madeno attempt to quantify post-Cambrian evolu-tion. They suggested ‘‘that a great amount ofbody-plan reorganization must have takenplace post Cambrian, to generate such distinc-tive taxa as for example the spiders and flyinginsects’’ (p. 259). However, the insects consid-ered in the analysis of Cambrian and Recentarthropod disparity performed by Wills et al.(1994) plotted closer to the global centroidthan did 16 of the Cambrian taxa (spiderswere omitted from the study). Likewise theliving Acanthopriapulus lies closer to the overallcentroid for fossil and living Priapulida thansix of the ten fossil taxa included in Wills’s(1998) analysis of this phylum, and furtherfrom it than the Carboniferous Priapulites,which is included in the crown group by Buddand Jensen. Budd and Jensen’s redefinition ofphylum in terms of stem and crown groups,though helpfully objective, does not improveour understanding of the Cambrian radiation.Phylogenies tell us about the probable se-

quence of events, but quantification of dispar-ity (Briggs et al. 1992a; Wills et al. 1994), theapproach inspired by Gould’s Wonderful Life,remains a better way of measuring amounts ofevolution.

It has been suggested that species might beidentified on the basis of a DNA ‘‘bar code,’’the sequence of single gene, but there is somedoubt that any single gene could resolve allanimal species (Pennisi 2003). A higher taxonmight be defined by a control gene, if one wereunique to a single clade. Hox genes control theexpression of structures such as limbs, so thattheir mutations may have a profound effect onthe phenotype of the adult organism. It istempting to correlate homeotic changes withthe diversification of segmentation patterns inarthropods, for example, the group that pro-vides the most diverse and disparate Cambri-an sample and has also received substantialattention from developmental biologists to-day. The main problem with evoking such amodel, however, is that abrupt (‘‘saltational’’)changes might result in a new organism thatis not integrated functionally (Budd 1999). Itis possible, therefore, that changes in the fea-tures controlled by Hox genes accumulated byincrements, and the control hierarchy was im-posed subsequently to ensure an increase inefficiency in building the body plan (but seeRonshaugen et al. 2002). This is the interpre-tation favored, for example, for the origin ofinsect wings (Carroll et al. 1995). Thus themost primitive members of a clade may nothave evolved the diagnostic control genes justas they also lack the morphological charactersthat identify the living members of a phylum.

How then to define a phylum? The totalgroup (Fig. 1B), i.e., all the members of a clade,both stem and crown, has been used by thosegenerating molecular phylogenies. The diffi-culty of separating the earliest representativesof the stem groups of two diverging phyladoes not negate the validity of the branchingpoint as the origin of the clade. Nor does it di-minish the usefulness of analyses of early fos-sil taxa that specify the suite of characters dis-played and where the taxa lie on the stemgroup. Trilobites may be extinct but even cla-distics can accommodate them in the Phylum Ar-thropoda. What we know of the fossils present


in the Lower Cambrian indicates that they hadalready diverged a considerable morphologi-cal distance from the ancestral conditionwhen phyla might have been confusingly sim-ilar and impossible to classify on morphologyalone. It remains a feasible research programto decide on which side of a fundamental di-chotomy a given fossil lies.

The Importance of Other CambrianKonservat-Lagerstatten

The interval since the publication of Won-derful Life (Gould 1989) has witnessed signifi-cant new research on the Cambrian, and inparticular many important discoveries fromthe Lower Cambrian of Chengjiang in south-west China, the Sirius Passet fauna of NorthGreenland, and the Upper Cambrian phos-phatized ‘‘Orsten’’ of southern Sweden andelsewhere. The number of genera describedfrom the Chengjiang fauna is now only slight-ly less than that from the Burgess Shale, anddiscoveries are still being made. Discoveries ofnew soft-bodied fossil occurrences naturallyextend the ranges of taxa without hard parts,owing to the taphonomic bias in the fossil re-cord against non-biomineralized taxa.

Our understanding of metazoan phylogenyhas advanced considerably since 1989, largelythrough molecular data. The cladogram (Fig.2) looked very different prior to the recogni-tion of the Lophotrochozoa and Ecdysozoa.Regardless of phylogenetic hypotheses, how-ever, new fossil discoveries have extended orprovided more convincing evidence for thestratigraphic range of taxa, and have reducedthe length of ghost ranges.

The first attempt to use cladistic methods tounravel the phylogeny of the Burgess Shale ar-thropods was made by Briggs and Whitting-ton in 1981, but this was before the wide-spread use of computer algorithms and wasbased on judgments about the significance ofcharacter attributes. At that time the few taxathat showed a suite of characters diagnostic ofone of the major living groups (particularlythe arrangement of head appendages) wereassigned accordingly, but the majority wereconsidered problematica. On this basis Cana-daspis was interpreted as a crustacean (Briggs1978) and Sanctacaris as a primitive sister

group of the chelicerates (Briggs and Collins1988). The remainder were the basis for the‘‘twenty unique designs of arthropod’’ rec-ognized by Gould (1989: p. 209).

The pivotal role claimed for the BurgessShale bivalved arthropod Canadaspis as theearliest unequivocal crustacean (Briggs 1978,1992a) was undermined by cladistic analysisusing a complete range of characters (Briggset al. 1992a; Wills et al. 1994, 1998). More-likely candidates were discovered among thetiny phosphatized arthropods from the Or-sten of the Alum Shale of Sweden (Walossek1999), but these are Upper Cambrian in age. Itappears that Canadaspis originated lower inthe arthropod stem group than the Orstenstem Crustacea. However, a fossil from Shrop-shire, England, reported by Siveter et al.(2001), extended the range of the phosphato-copid crustaceans to the Lower Cambrian,confirming that the Eucrustacea had originat-ed by this time (the phosphatocopids are a sis-ter taxon to the Eucrustacea sensu Walossek1999). Cladistic analyses (Briggs et al. 1992a;Wills et al. 1994, 1998) placed Sanctacaris with-in the Arachnomorpha (Wills et al. 1998), butnot in proximity to the living chelicerates.However, the recent description of a larvalpycnogonid from the Swedish Orsten pro-vides evidence for a Cambrian origin of chel-icerates (Waloszek and Dunlop 2002). (Walo-szek and Dunlop [2002] considered pycnogo-nids to form a sister group to the euchelicer-ates, but Giribet et al. [2001] placed them as asister group to all other arthropods.)

Gould (1989: p. 171) considered that Whit-tington was incorrect to regard Aysheaia as arepresentative of a separate group, consider-ing that it ‘‘should be retained among the On-ychophora.’’ In this respect Gould appears tohave been correct—the discovery of severalCambrian lobopods, particularly in the Che-ngjiang biotas, has revealed a diverse groupthat includes the modern onychophorans(Ramskold and Hou 1991; Ramskold andChen 1998). Ironically, of course, Hallucigeniaalso belongs here (Ramskold and Hou 1991),although the incomplete information availableto Conway Morris (1977a), and consequentlyGould, prevented this realization.

Ramskold and Chen (1998) included living


FIGURE 2. The fossil record and metazoan phylogeny. Relationships are illustrated as in Pennisi 2003. Dark linesrepresent the stratigraphic range of major metazoan taxa from their first appearance in the fossil record, indicatedby a square, to the present. Light lines represent ghost ranges implied by a previous occurrence of a related taxon.The data on Lobopodia, Brachiopoda, Mollusca, Hemichordata, and Echinodermata, as well as phyla that first ap-pear after the Cambrian, are from Benton 1993. Other first appearances are discussed in the text. Since WonderfulLife (1989), when first appearances were as indicated here by circles, discoveries in the Chengjiang fauna havepushed various Burgess Shale occurrences back 20 million years—Ctenophora, Lobopodia, Priapulida, Chaetog-natha, Cephalochordata—and the vertebrates back from the Ordovician. The range of the tardigrades has extendedfrom the Tertiary to the Cambrian, and the tunicates have acquired a reliable Cambrian record. Sponges have beendiscovered in the Vendian.

Onychophora (a single datum) and the Car-boniferous taxon Ilyodes in an analysis withthe Cambrian lobopods. Their strict consen-sus revealed the living Onychophora and Ily-odes as a clade in a polychotomy with twoclades of Cambrian lobopodians (one contain-ing Hallucigenia). In three of their nine solu-tions, however, the clade that includes the Re-cent forms is a sister group to one includingall the Cambrian lobopodians. These solutionsplace all the Cambrian lobopodians in thecrown group onychophorans and supportGould’s view that Aysheaia does indeed belonghere. In any event Ramskold and Chen’s (1998)analysis shows that at least one of the twoclades of fossil lobopods must be a sistergroup of modern onychophorans and lie with-in the crown group. Budd and Jensen (2000: p.

259) regarded Aysheaia as ‘‘probably best con-sidered not to lie in even the onychophoranstem lineage,’’ but this assertion was not sup-ported by a parsimony analysis (Budd 1996).The place of the tardigrades in arthropod evo-lution remains controversial. Dewel and Dew-el (1998) placed them in the arthropod stemgroup between lobopods and anomalocari-dids but conceded that this placement wasconjectural. Importantly, however, Muller etal. (1995) described a stem-group tardigradefrom the Middle Cambrian of Siberia extend-ing the origins of the group back to the Cam-brian.

Gould illustrated his Wonderful Life thesiswith Pikaia, a laterally compressed, ribbon-shaped animal that displays the chordate fea-tures of myomeres and a notochord, but bears


an unusual pair of tentacles on the ‘‘head.’’Gould argued that humans exist ‘‘because Pi-kaia survived the Burgess decimation’’ (1989:p. 323). Although he emphasized (p. 322) thathe would not ‘‘be foolish enough to state thatall opportunity for a chordate future residedwith Pikaia in the Middle Cambrian,’’ the factthat Pikaia was then the only chordate knownfrom the Burgess Shale added a certain pi-quancy to the story and made the argumentmore easily accessible. Pikaia is now regardedas representing ‘‘a grade comparable to ceph-alochordates’’ (Shu et al. 1999: p. 46) and notdirectly ancestral to the vertebrates (Shu 2003:p. 733).

Gould doubted (1989: p. 149) that cono-donts should be assigned to the chordates:Wonderful Life was written before evidence fortheir chordate affinities (Aldridge et al. 1986)was widely accepted by paleontologists (seeBriggs 1992b; Ahlberg 2001). Gould (p. 322)believed that ‘‘fossils of true vertebrates, ini-tially represented by agnathan, or jawless,fishes, first appear in the Middle Ordovician. . . .’’ He predicted (p. 322), however, that‘‘other chordates, as yet undiscovered, musthave inhabited Cambrian seas,’’ a prophesyfulfilled by the subsequent discovery of twonew poorly known forms from the BurgessShale (Simonetta and Insom 1993; Smith et al.2001) and, more importantly, by a remarkablediversity in the Chengjiang biota. Arthropodsare diverse in both Burgess and Chengjiangbiotas, but chordates are much more abundantin the Chinese material.

Haikouichthys (Fig. 3A) (Shu et al. 1999,2003a) is fishlike, with well-documented gen-eralized vertebrate characters, and it has beenaccepted rapidly as an early craniate. Thehead, with large eyes and paired nostrils, andthe apparent vertebral elements (Shu et al.2003a) demonstrate its craniate affinities—and indicate that vertebrate evolution was welladvanced by the Early Cambrian. Shu et al.(1999) originally placed Haikouichthys directlycrownward of hagfishes. However, recent mo-lecular-phylogenetic studies indicate that hag-fishes are not stem craniates, as previouslythought, but are highly derived vertebratesthat group with lampreys as cyclostomes(Takezaki et al. 2003; Mallatt and Sullivan

1998). This leaves Haikouichthys as a stem-group craniate (Shu et al. 2003a). Myllokun-mingia (Shu et al. 1999) is similar to Hai-kouichthys and has been interpreted as a seniorsynonym (Hou et al. 2002; but see ConwayMorris 2003b). An additional Chengjiang ag-nathan, Zhongjianichthys (Shu 2003), is alsosimilar, suggesting that there was not muchmorphological diversity among these EarlyCambrian vertebrates. Hard parts first appearin the Middle Cambrian Euconodonta.

Vetulicola (Fig. 3B), a large Chengjiang ani-mal with a carapace-like anterior part andsegmented tail-like posterior part, was firstdescribed as a large bivalved arthropod (Hou1987). Chen and Zhou (1997) commented onits enigmatic nature and assigned it to a newclass of stem-group arthropods, Vetulicolida.They argued that Banffia from the BurgessShale belongs to the same group. A new taxonDidazoon was described by Shu, Conway Mor-ris, and others (Shu et al. 2001b) and it, to-gether with additional data on Xidazoon andVetulicola, prompted them to erect a new phy-lum Vetulicolia. This should have delightedGould—after all Wonderful Life was based ondiscoveries of new Cambrian phyla and here,over ten years later, examples were still turn-ing up. The interpretation of perforations inthe anterior body region as precursors of gillslits led Shu et al. (2001b, 2003b) to regard ve-tulicolians as primitive deuterostomes. Gee(2001), however, remarked on the similarity ofthe vetulicolian body plan to that of a tunicatetadpole larva and suggested that they may oc-cupy a more crownward position and be thesister group of chordates. Lacalli (2002) andJefferies (personal communication 2003) con-sidered the possibility that vetulicolians arestem tunicates. In contrast, Butterfield (2003)and Mallatt et al. (2003) argued that Vetulicolahas a cuticle and is therefore more likely tohave been an arthropod. A tunicate was de-scribed from the Chengjiang Lagerstatte byShu et al. (2001a). Doubts were cast on the tu-nicate nature of this fossil by Chen et al.(2003), who recognized additional specimenswith large lophophore-like tentacles. They de-scribed another animal from the Lower Cam-brian of China as a tunicate, on the basis of alarger number of specimens. New phyla of


FIGURE 3. Chordates and other animals from the Lower Cambrian Chengjiang biota of Yunnan Province, southernChina. A, Haikouichthys ercaicunensis, HZ-f-12-127 (Yunnan Institute of Geological Sciences, Kunming), 33.85. B,Vetulicola cuneata, ELRC 19397a (Early Life Research Center, Chengjiang), 31.2. C, Yunnanozoon lividum, ELRC52004, 35.49. D, Haikouella lanceolatum, ELRC 00258, 35.23. D, Possible chaetognath Eognathacantha ercainella, ELRC02001a, 33.73. Photographs kindly provided by J.-Y. Chen (Fig. 3A with the permission of L.-J. Chen, Yunnan In-stitute of Geology).


modern metazoans are described rarely, andwhen they are, such a designation usually re-flects a very simple morphology with a lack ofdiagnostic features. Their affinities can betested by using molecular and embryologicalevidence (Winnepenninckx et al. 1998 on Cy-cliophora; Bourlat et al. 2003 on Xenoturbella;Ruiz-Trillo et al. 2002 on Acoela). New phy-lum status is not proposed often for fossils ei-ther. Given the need to rely on morphology, al-lied to the vagaries of taphonomy and the dif-ficulty of interpreting unfamiliar form, theerection of new extinct phyla is fraught withdifficulty. At the very least it demands explicithypotheses of character distribution and ananalysis of relationships within a cladisticcontext.

Yunnanozoon (Fig. 3C), a blade-shaped, soft-bodied, segmented animal from the Che-ngjiang fossil fauna, was first described byHou et al. (1991) as a wormlike problemati-cum. Chen et al. (1995) described additionalmaterial, including evidence of a notochordand myomeres, and interpreted Yunnanozoonas a cephalochordate. They noted that this pre-dicted that ‘‘other chordate clades (tunicatesand craniates) had evolved by the Late Atda-banian in the main burst of the Cambrian Ex-plosion’’ (p. 720). Dzik (1995) considered thatYunnanozoon ‘‘belongs to a completely extinctgroup of the earliest chordates’’ (p. 352). Heerected a Class Yunnanozoa, Order Yunna-nozoida, Family Yunnanozoidae. A problemwith the chordate interpretation of the yun-nanozoans is the unusual ventral position ofthe notochord relative to the myomeres. Shuet al. (1996b) collected additional material andreinterpreted Yunnanozoon as the earliestknown hemichordate, on the basis of their in-terpretation of features as proboscis, collarand trunk. Chen and Li (1997) redescribedYunnanozoon and reinterpreted it as a primi-tive chordate, arguing that the hemichordatefeatures are artifacts of decay. They also con-sidered that the single specimen described byShu et al. (1996a) as Cathaymyrus, a Pikaia-likechordate, is, in fact, an example of Yunnano-zoon. Chen et al. (1999) discovered abundantwell-preserved specimens of a new yunnano-zoan, which they described as Haikouella (Fig.3D), that appeared more craniate-like than

cephalochordate-like. Holland and Chen(2001) reviewed the evidence for the originand early evolution of the vertebrates anddemonstrated, by means of a cladistic analy-sis, that yunnanozoans are the sister group ofthe vertebrates, lying crownward of the ceph-alochordates. Shu et al. (2003b) described anew species of Haikouella, adding new infor-mation on the morphology of the yunnano-zoans, including external gills. They arguedthat there is no evidence for a chordate affinity,but that these animals were stem-group deu-terostomes, allied to the Phylum Vetulicolia,to which they tentatively assigned them. Mal-latt et al. (2003) discussed this paper and re-asserted the craniate affinity for yunnano-zoans based on the presence of gills and evi-dence for muscle fibers in the myomeres, aclaim that Shu and Conway Morris (2003) re-jected.

Mallatt and Chen (2003) provided a test ofthe hypothesis. They demonstrated the cleardifference between cuticular folds and musclefibers in Haikouella, illustrating strong evi-dence for the latter. More importantly theypresented a detailed description and justifi-cation of a matrix of 40 characters that theyused to perform a cladistic analysis of the re-lationships of Haikouella (i.e., yunnanozoans),protostomes, hemichordates, tunicates, ceph-alochordates, conodonts, and craniates. Theresult confirmed that of Holland and Chen(2001): Haikouella, and therefore the Yunna-nozoa, is a sister group of vertebrates. A sim-ilar cladistic analysis to determine the posi-tion of vetulicolians (deuterostomes, arthro-pods, or other) remains to be done. Not to doso, once the evidence can be assembled, is toignore the lessons of the debate that WonderfulLife stimulated.

Several new discoveries since 1989 in theLower Cambrian Chengjiang fauna of south-ern China have extended the range of othergroups that are also represented in the MiddleCambrian Burgess Shale. Two new genera ofCtenophora were described by Chen andZhou (1997). A likely chaetognath from theBurgess Shale discovered by D. Collins (RoyalOntario Museum) awaits description (Briggsand Conway Morris 1986: p. 167). In the mean-time a possible chaetognath (Fig. 3E) from


Kunming was reported by Chen and Huang(2002), extending the record of this groupfrom the Carboniferous to the Lower Cambri-an. The Phoronida are represented by a singlegenus from Chengjiang (Chen and Zhou1997), the only body fossil of this groupknown. Sponges were reported from the Edi-acara fauna of South Australia (Gehling andRigby 1996).

Divergence Times of Major Clades and theCambrian Explosion

The Chengjiang fauna (Atdabanian) dem-onstrates that, among fossilizable clades, sev-eral phylum-or class-level taxa extended backinto the Early Cambrian, even allowing for thecaveats about definitions of higher taxa dis-cussed above. It is inescapable that the diver-gence of these groups from their common an-cestors happened before the time representedby this early diverse fauna (Fortey et al. 1996).In spite of the discovery of earliest Cambrian(Nemakyt–Daldyn) embryos (Bengtson andYue 1997) it remains debatable whether thesederived groups were also present during theduration of this stage (ca. 10 Ma)—althoughthe presence of plausible arthropod and mol-lusc trace fossils (Brasier et al. 1994) perhapscould be taken as evidence that some of themajor groups had already appeared. The sug-gestion that the absence of Precambrian meta-zoan fossils might be attributed to small size(Fortey et al. 1996) has proved controversial.Budd and Jensen (2000) inferred that the an-cestral bilaterian would have been large, butRuiz-Trillo et al. (2002) have argued that theancestor was a small, simple, probably direct-developing acoel, judging from a phylogeneticanalysis of platyhelminthes that showed themto be polyphyletic (see Peterson et al. 2005).

The affinities of the Vendian Ediacaran or-ganisms remain controversial. They have longbeen interpreted as ‘‘ancestral’’ or stem-groupmetazoans (e.g., Glaessner 1984; Gehling1991). Gould (1989), however, favored the al-ternative interpretation (Seilacher 1989) of theEdiacaran organisms as the product of a Pre-cambrian evolutionary experiment that gen-erated unfamiliar creatures that existed along-side metazoans, the latter represented mainlyby trace fossils (e.g., those that show burrow-

ing by peristaltic locomotion). In followingSeilacher’s interpretation Gould (1989) em-phasized the evolutionary importance of theCambrian explosion—the fossil evidence in-dicates that some phyla, such as brachiopodsand arthropods, may not have evolved by latePrecambrian times. The affinities of the Edi-acaran organisms continue to excite contro-versy. Seilacher extended his interpretation(Seilacher et al. 2003) to argue that many of theproblematic Ediacaran organisms representprocaryotic biomats and giant rhizopodanprotozoans (Xenophyophoria and Vendobion-ta) that dominated a late Precambrian worldin which metazoans were rare (sponges, coe-lenterates, a mollusc, and an echinoderm) andrelatively small. This view contrasts markedlywith that of Fedonkin (2003), who regardedmost of the Ediacaran animals as diploblasticor triploblastic metazoans. Important evi-dence for Fedonkin’s interpretation is the dis-covery of several trace fossils associated withthe Ediacaran organisms that made them, in-cluding Yorgia, Kimberella, and Dickinsonia.

Given that there is no consensus on the af-finities of the Ediacaran body fossils, whichare relatively abundant and well preserved, itis not surprising that the older trace fossil ev-idence for metazoan activities has provedeven more controversial. Early records havebeen scrutinized carefully as a potentialmeans of reconciling dates for the origin ofgroups based on molecular data with the ev-idence of the fossil record. Budd and Jensen(2000) argued, however, that none of the tracefossils reported to predate the last Proterozoicice age (Marinoan or Varangerian) stand up toscrutiny—they can be attributed to sedimen-tary structures or metaphytes, or they havebeen misdated. The difficulty here is that ofpreservation and distortion over time. The ear-liest trace fossils that are attributed confident-ly to stem-group bilaterians are about 555 mil-lion years old (Droser et al. 2002).

Unequivocal fossils of identifiable metazo-ans older than the base of the Cambrian havenot been forthcoming. The embryos from theNeoproterozoic Doushantuo Formation arepreserved at too early a stage to allow them tobe identified to specific metazoan taxa (Xiaoand Knoll 2000). Thus dating of molecular


trees is based inevitably on Cambrian fossiloccurrences. At the time Gould wrote Wonder-ful Life, methods of determining from inde-pendent evidence the timing of the deeperbranches of phylogeny were in their infancy—as, for example, the divergence between chor-dates and echinoderms, or between proto-stomes and deuterostomes. Estimates becamea possibility as comparisons of genomic se-quences, such as that of small subunit RNA,from a variety of organisms allowed models tobe developed for converting sequence differ-ence into a measure of time, using assump-tions about the ‘‘molecular clock’’ (Smith andPeterson 2002).

A pioneering study by Wray et al. (1996)used data from 18S ribosomal RNA and sevengenes to estimate divergence times betweenmetazoan phyla. Divergence times in the deepPrecambrian were inferred: for example about1200 Ma for a protostome-deuterostome split,and 1000 Ma for the echinoderm-chordatesplit. Thus, according to this estimate, phylo-genetic divergence long preceded the Cam-brian ‘‘explosion.’’ Other analyses soon fol-lowed. Wang et al. (1998), for example, iden-tified the chordate arthropod split at 995 6 79Ma and noted that ‘‘basal animal phyla (Por-ifera, Cnidaria, Ctenophora) diverged be-tween 1200–1500 Ma,’’ and ‘‘six animal phylaoriginated deep in the Precambrian more than400 Ma earlier than their first appearance inthe fossil record.’’ Ayala et al. (1998) usedpairwise distance methods to estimate diver-gence times, and they obtained a proterosto-me-deuterostome split at 670 Ma and an echi-noderm chordate split at 600 Ma. Hence with-in a few years it was already apparent that us-ing different genes and different methodsyielded strikingly different estimates of diver-gence times between the same clades.

Methodological limitations may have beenresponsible for this range in estimates of di-vergence times. Confidence limits (e.g., inWang et al. 1999) were often calculated by tak-ing the standard error of the mean divergenceestimates as the total estimate error, therebyignoring the assumed Poisson variance of thesubstitutions. Bromham et al. (1998) noted thedifficulties of incorporating this variance intothe calculations, which concatenate the error

on estimates progressively with deeper diver-gences; they used a method based on maxi-mum likelihood. This increased the length ofconfidence limits on divergence times consid-erably compared with those derived by usingother methods; although the means of the pro-terostome-deuterostome and echinoderm-chordate splits were still well within the Pre-cambrian, for example, the upper bounds onthe confidence limits of the latter approachedthe lower Cambrian. Further problems with‘‘molecular clock’’ estimations of divergencetimes were outlined by Cutler (2000), whopointed out that usual approaches did nottake into account ‘‘overdispersed’’ loci—thosethat fail relative rate tests—and that the Pois-son assumptions tend to underestimate theconfidence limits for such loci. He developeda maximum likelihood method using moregeneral stationary process assumptions, andwas able to show that the Poisson-based mod-el could be rejected for some loci. Cutler’smethod produced considerable error bars ondeep divergences, but the origin of metazoansclose to the base of the Cambrian could not beruled out, although it was less likely than asignificantly older divergence.

Methods for ‘‘relaxing’’ the molecular clock,allowing for different rates in different partsof the tree, have been developed in the last fewyears and are likely to have a positive impacton accuracy of estimates of divergence times(Sanderson 2002). Bayesian approaches (Huel-senbeck et al. 2001), using Markov ChainMonte Carlo methods, have become compu-tationally feasible, although these methodshave not escaped criticism (Shoemaker et al.1999). At the same time, the comparative anal-ysis of a wider range of single-copy nucleargenes is likely to identify more reliable loci forthe analysis of deep divergences and will bepreferable to the earlier reliance upon 18S ormitochondrial genes. Sequencing of a criticalrange of invertebrate organisms selected ac-cording to their phylogenetic position (Smithand Peterson 2002)—rather than their readyavailability as laboratory material—shouldhelp tackle some of the long-branch problemsthat have also influenced the accuracy of pub-lished divergence estimates.

As yet there is no ‘‘definitive’’ answer to the


question regarding the length of the Precam-brian phylogenetic fuse(s). If one trend is dis-cernible in modern studies of divergencetimes, it is that the earlier papers tended toproduce very deep dates that may, in fact, bepartly artifacts. As methods have becomemore refined and critical, mean dates for di-vergences have become younger, but errorranges have also generally become longer.Nonetheless, some of the fundamental‘‘splits’’ in the tree of life fall consistently wellwithin the Precambrian, and it would, per-haps, be surprising if they were to ‘‘jump’’considerably upwards as a result of futurework. Within the Metazoa, for example, theproterostome-deuterostome divergence in themajority of studies falls within the range 750–1000 Ma (Fortey et al. 2003: Fig. 3.3)—the ex-ception is the latest Precambrian date of Lynch(1999). This suggests a 200 Ma period to allowdiversification in metazoan phyla before thebase of the Cambrian. However, there is stillmuch flexibility in the estimates of divergencetimes between pairs of phyla. For example,Hedges and Kumar (2003) suggest a diver-gence between arthropods and vertebrates at993 Ma and vertebrates from hemichordates at751 Ma (error ranges ignored). However, Pe-terson et al. (2004, 2005) suggested that pre-vious molecular estimates of divergence timesmay have been influenced by calibrating onthe basis of rates of genetic change in verte-brates, which are slower than those in inver-tebrates, a difference that would push backtimes of origin artificially. They estimated thelast common ancestor of protostomes anddeuterostomes (exclusive of acoelomorphs) atca. 570 Ma and the last common ancestor ofprotostomes at ca. 560 Ma. This opens up thepossibility again that there was, indeed, a pe-riod of ‘‘explosive’’ origination—a return, ineffect, to Gould’s original formulation (Con-way Morris 2000). However, it is still earlydays in the methodology: for example, onemight ask even whether rates in invertebratesare ‘‘one size fits all’’?

The question of molecular dating of phyleticdivergences remains a rapidly changing field,and it may be premature to prejudge the out-come as it relates to the base of the Cambrian.At this stage, it may be safe to conclude that

the basal divisions, as between Ecdysozoa,protostomes, and deuterostomes, were ac-complished well within the Precambrian andprovide a long phylogenetic ‘‘fuse.’’ However,the timing of origin of the phyla that appearedat, or after, the basal Cambrian is still not firm-ly established and varies widely according tomethodology and clock assumptions. Takingmolecular grounds alone, Gould’s ‘‘explosive’’scenario is still a feasible option. For example,a recent application of Bayesian methods (ArisBrosou and Yang 2003) using 22 genes, andpermitting rates to vary in time, identified amajor burst of molecular evolution late in Pre-cambrian time, bringing molecular estimatesof metazoan divergence closer to the base ofthe Cambrian than in all previous studies.

Contingency

Perhaps the most lasting image in WonderfulLife is the device that Gould used to explainthe concept of contingency: ‘‘Wind the tape oflife back to Burgess times, and let it playagain. If Pikaia does not survive in the replay,we are wiped out of future history—all of us,from shark to robin to orangutan’’ (Gould1989: p. 323). There is no doubt that contin-gency is an important factor in evolution, notleast to the extent that the loss of many taxaduring mass extinctions, particularly those at-tributed to asteroid impacts, is randomly de-termined. However, Conway Morris (1998,2003a), in particular, has argued that conver-gent evolution lends an inevitability to theemergence of a creature with faculties similarto our own. The two antagonists summarizedtheir viewpoints in Natural History (ConwayMorris and Gould 1998). Convergence on thisscale, however, awaits serious investigation.The degree of convergence within groups oforganisms can by quantified by using parsi-mony programs (in terms of homoplasy).Mapping genomes will eventually determinehow deep in the tree of life control genes con-strain and determine potential outcomes.Computer simulations can determine theprobability of sequences of events. In themeantime, to recast the history of life in termsof convergence places a singular emphasis onone aspect of a complex story. The repeatedevolution of sabertooth morphology among


predatory mammals is doubtless an extraor-dinary phenomenon that might suggest somesort of tendency to evolve in a particular di-rection. However, the fossil record has also re-vealed designs that have never been replicat-ed. Graptolites, for example, colonial plank-tonic organisms with an extraordinary arrayof spiral, s-shaped, or branching rhabdo-somes, have never had close homeomorphs inthe 400 million years since their extinction. Ishumankind a ‘‘sabertooth’’ or a ‘‘graptolite’’?Unfortunately the probability that conscious-ness would have evolved even if the verte-brates had been eliminated by some late Cam-brian extinction is unlikely to be susceptible toscientific enquiry.

Conclusions

Gould’s Wonderful Life (1989) was an inspi-ration to students and served to set the agendafor 15 years of intensive research on the mean-ing of the Cambrian evolutionary ‘‘explo-sion.’’ New sources of data have arisen fromseveral quarters—particularly the discoveryof major new soft-bodied faunas from the Pa-leozoic and the application of molecular tech-niques to questions of phylogeny and diver-gence times. Many of the issues that Gouldraised are still under debate—a number haveattracted the attention of zoologists, paleon-tologists, and molecular biologists and fos-tered interdisciplinary interest in a way thatshould be a model for the future. Althoughsome of the ideas in Wonderful Life have notstood the test of time, the challenges set haveserved to move the science forward.

Claims about strikingly elevated numbersof ‘‘new phyla’’ in the Cambrian have beencountered. Cladistic analysis has placed themajority of Cambrian ‘‘weird wonders’’ asstem taxa of known clades. New fossil discov-eries have resolved the problematic status ofseveral former enigmas. Although new majorextinct taxa (e.g., Phylum Vetulicolia [Shu etal. 2001b]) occasionally still appear in the lit-erature, their erection is controversial andtheir objective basis unclear. Nor have claimsabout a maximum of morphological disparityin the Cambrian stood up to analysis. At most,disparity of design was equal to that at the Re-cent. Its range was established early and its

compass shifted through time in total mor-phospace.

Debates about the Cambrian radiation havestimulated a debate over how ‘‘phyla’’ mightbe defined at all. One view bases a definitionof phylum (or class) on the crown group andexcludes stem taxa as plesions. This has theadvantage of objectivity and ensures that mo-lecular data are available for the whole taxonexcept where primitive survivors pull fossiltaxa into the crown clade. The crown-groupdefinition of a phylum is arbitrary, however,depending on the most primitive taxon thathappens to have survived to the present day.An alternative view defines higher taxa on thebasal divergence (total group), which is logi-cally consistent, and assigns stem taxa to aphylum. A disadvantage is that the placing ofextremely plesiomorphic forms, particularlythose with peculiar autapomorphies, remainsdifficult.

The question of the length of the phyloge-netic ‘‘fuse’’ below the Cambrian—the gener-ation time of the living phyla—remains un-resolved. Molecular-based estimates of diver-gence times vary widely according to themethods used and are only now being sup-plied with realistic error ranges. Nonetheless,deep Precambrian divergence of ‘‘superphy-la’’ (protostomes, Ecdysozoa, deuterostomes)is favored by the majority of studies. It istempting to relate subsequent rapid evolution-ary events in the origin of phyla to global sce-narios of environmental change, such as‘‘snowball earth’’ (Hoffman et al. 1998; see Pe-terson et al. 2005). Such an attractive hypoth-esis must not, however, discourage efforts toaugment the fossil record in the late Protero-zoic, because one good fossil at an unexpectedtime or place could still have the effect of over-turning any preconceptions.

Acknowledgments

We are grateful to S. Bengtson, J. Mallatt,and F. R. Schram for comments on an earlierdraft of the manuscript. We acknowledge themajor contribution of M. A. Wills to enablingus to address some of the major issues raisedin Wonderful Life. J.-Y. Chen and M. A. Willsprovided illustrations, and S. Butts assistedwith the preparation of figures.


Literature Cited

Ahlberg, P. E., ed. 2001. Major events in early vertebrate evo-lution. Systematics Association Special Volume 61. Taylor andFrancis, London.

Aldridge, R. J., D. E. G. Briggs, E. N. K. Clarkson, and M. P.Smith. 1986. The affinities of conodonts—new evidence fromthe Carboniferous of Edinburgh, Scotland. Lethaia 19:279–291.

Allison, P. A., and D. E. G. Briggs. 1993. Exceptional fossil re-cord: distribution of soft-tissue preservation through thePhanerozoic. Geology 21:527–530.

Aris-Brosou, S., and Z-H. Yang. 2003. Bayesian models of epi-sodic evolution support a late Precambrian explosive diver-gence of the Metazoa. Molecular Biology and Evolution 20:1947–1954.

Averof, M., and M. Akam. 1995. Insect-crustacean relationships:insights from comparative developmental and molecularstudies. Philosophical Transactions of the Royal Society ofLondon B 347:293–303.

Ayala, F. J., A. Rzhetsky, and F. J. Ayala. 1998. Origin of themetazoan phyla: molecular clocks confirm paleontological es-timates. Proceedings of the National Academy of SciencesUSA 95:606–12.

Bengtson, S., and Z. Yue. 1997. Fossilized metazoan embryosfrom the earliest Cambrian. Science 277:1645–1648.

Benton, M. J., ed. 1993. The fossil record 2. Chapman and Hall,London.

———. 2000. Stems, nodes, crown clades, and rank-free lists: isLinnaeus dead? Biological Reviews 75:633–648.

Bourlat, S. J., C. Nielsen, A. E. Lockyer, D. T. J. Littlewood, andM. J. Telford. 2003. Xenoturbella is a deuterostome that eatsmolluscs. Nature 424:925–928.

Brasier, M. J., J. W. Cowie, and M. E. Taylor. 1994. Decision onthe Precambrian–Cambrian boundary. Episodes 17:3–8.

Briggs, D. E. G. 1978. The morphology, mode of life and affini-ties of Canadaspis perfecta (Crustacea: Phyllocarida), MiddleCambrian, Burgess Shale. Philosophical Transactions of theRoyal Society of London B 281:439–487.

———. 1992a. Phylogenetic significance of the Burgess Shalecrustacean Canadaspis. Acta Zoologica, Stockholm 73:293–300.

———. 1992b. Conodonts—a major extinct group added to thevertebrates. Science 256:1285–1286.

Briggs, D. E. G., and C. Bartels. 2001. New arthropods from theLower Devonian Hunsruck Slate (Lower Emsian, RhenishMassif, western Germany). Palaeontology 44:275–303.

Briggs, D. E. G., and D. Collins. 1988. A Middle Cambrian chel-icerate from Mount Stephen, British Columbia. Palaeontology31:779–798.

Briggs, D. E. G., and S. Conway Morris. 1986. Problematica fromthe Middle Cambrian Burgess Shale of British Columbia. Pp.167–183 in A. Hoffman and M. H. Nitecki, eds. Problematicfossil taxa. Oxford University Press, Oxford.

Briggs, D. E. G., and R. A. Fortey. 1989. The early radiation andrelationships of the major arthropod groups. Science 246:241–243.

Briggs, D. E. G., and H. B. Whittington. 1981. Relationships ofarthropods from the Burgess Shale and other Cambrian se-quences. Proceedings of the second International Symposiumon the Cambrian System. U.S. Geological Survey Open-fileReport 81-743:38–41.

Briggs, D. E. G., R. A. Fortey, and M. A. Wills. 1992a. Morpho-logical disparity in the Cambrian. Science 256:1670–1673.

———. 1992b. Cambrian and Recent morphological disparity:response to Foote and Gould, and Lees. Science 258:1817–1818.

Briggs, D. E. G., D. J. Siveter, and D. J. Siveter. 1996. Soft-bodied

fossils from a Silurian volcaniclastic deposit. Nature 382:248–250.

Bromham, L., A. Rambaut, R. A. Fortey, A. Cooper, and D. Pen-ny. 1998. Testing the Cambrian explosion hypothesis using amolecular dating technique. Proceedings of the NationalAcademy of Sciences USA 95:12386–9.

Budd, G. E. 1993. A Cambrian gilled lobopod from Greenland.Nature 364:709–711.

———. 1996. The morphology of Opabinia regalis and the recon-struction of the arthropod stem-group. Lethaia 29:1–14.

———. 1999. Does evolution in body patterning genes drivemorphological change—or vice versa? BioEssays 21:326–332.

———. 2002. A palaeontological solution to the arthropod headproblem. Nature 417:271–75.

———. 2003. The Cambrian fossil record and the origin of thephyla. Integrative and Comparative Biology 43:157–165.

Budd, G. E., and S. Jensen. 2000. A critical appraisal of the fossilrecord of the bilaterian phyla. Biological Reviews 75:253–295.

Butterfield, N. J. 2003. Exceptional fossil preservation and theCambrian explosion. Integrative and Comparative Biology43:166–177.

Carroll, S. B., S. Weatherbee, and J. Langeland. 1995. Homeoticgenes and the regulation and evolution of insect wing num-ber. Nature 375:58–61.

Chen, J.-Y., and D.-Y. Huang. 2002. A possible Lower Cambrianchaetognath (arrow worm). Science 298:187.

Chen, J.-Y., and C.-W. Li. 1997. Early Cambrian chordate fromChengjiang, China. Bulletin of the National Museum of Nat-ural Science 10:257–273.

Chen, J.-Y., and G.-Q. Zhou. 1997. Biology of the Chengjiang fau-na. Bulletin of the National Museum of Natural Science 10:11–105.

Chen, J.-Y., X.-G. Hou, and G.-X. Li. 1989. Early Cambrian nettedscale-bearing worm-like sea animal. Acta PalaeontologicaSinica 29:402–414.

Chen, J.-Y., J. Dzik, G. E. Edgecombe, L. Ranskold, and G.-Q.Zhou. 1995. A possible Early Cambrian chordate. Nature 377:720–722.

Chen, J.-Y., D.-Y. Huang, and C.-W. Li. 1999. An early Cambriancraniate-like chordate. Nature 402:518–522.

Chen, J.-Y., D.-Y. Huang, Q.-Q. Peng, H.-M. Chi, X.-Q. Wang,and M. Feng. 2003. The first tunicate from the Early Cambrianof South China. Proceedings of the National Academy of Sci-ences USA 100:8314–8318.

Collins, D. 1996. The ‘‘evolution’’ of Anomalocaris and its clas-sification in the arthropod class Dinocarida (nov.) and orderRadiodonta (nov.). Journal of Paleontology 70:280–93.

Conway Morris, S. 1977a. A new metazoan from the CambrianBurgess Shale, British Columbia. Palaeontology 20:623–640.

———. 1977b. Fossil priapulid worms. Special Papers in Pa-laeontology 20:1–98.

———. 1998. The crucible of creation: the Burgess Shale and therise of animals. Oxford University Press, Oxford.

———. 2000. The Cambrian ‘‘explosion’’: slow fuse or megaton-nage? Proceedings of the National Academy of Sciences USA97:4426–9.

———. 2003a. Life’s solution: inevitable humans in a lonely uni-verse. Cambridge University Press, Cambridge.

———. 2003b. The Cambrian ‘‘explosion’’ of metazoans and mo-lecular biology: would Darwin be satisfied? InternationalJournal of Developmental Biology 47:505–515.

Conway Morris, S., and S. J. Gould. 1998. Showdown on the Bur-gess Shale. [The Challenge by Simon Conway Morris and theReply by Stephen Jay Gould.] Natural History 107:48–55.

Conway Morris, S., and J. S. Peel. 1995. Articulated halkieriidsfrom the Lower Cambrian of North Greenland and their rolein early protostome evolution. Philosophical Transactions ofthe Royal Society of London B 347:304–358.


Cooper, A., and R. A. Fortey. 1998. Evolutionary explosions andthe phylogenetic fuse. Trends in Ecology and Evolution 13:151–6.

Cutler, D. J. 2000. Estimating divergence times in the presenceof an overdispersed molecular clock. Molecular Biology andEvolution 17:1647–60.

Dewel, R. A., and W. C. Dewel. 1998. The place of tardigradesin arthropod evolution. Pp. 109–123 in R. A. Fortey and R. H.Thomas, eds. Arthropod relationships. Chapman and Hall,London.

Droser, M. L., S. Jensen, and J. G. Gehling. 2002. Trace fossilsand substrates of the terminal Proterozoic-Cambrian transi-tion: implications for the record of early bilaterians and sed-iment mixing. Proceedings of the National Academy of Sci-ences USA 99:12572–12576.

Dzik, J. 1995. Yunnanozoon and the ancestry of the chordates.Acta Palaeontologica Polonica 40:341–360.

Edgecombe, G. D., ed. 1998. Arthropod fossils and phylogeny.Columbia University Press, New York.

Fedonkin, M. A. 2003. The origin of the Metazoa in the light ofthe Proterozoic fossil record. Paleontological Research 7:9–41.

Foote, M., and S. J. Gould. 1992. Cambrian and Recent morpho-logical disparity. Science 258:1816.

Fortey, R. A. 1989. The collection connection. [Review of Won-derful Life.] Nature 342:303.

Fortey, R. A., D. E. G. Briggs, and M. A. Wills. 1996. The Cam-brian evolutionary ‘explosion’: decoupling cladogenesis frommorphological disparity. Biological Journal of the Linnean So-ciety 57:13–33.

Fortey, R. A., J. Jackson, and J. Strugnell. 2003. Phylogenetic‘‘fuses’’ and evolutionary explosions: conflicting evidenceand critical tests. Pp. 41–65 in P. Donoghue, ed. Molecular evo-lution. Chapman and Hall.

Gee, H. 2001. On being vetulicolian. Nature 414:407–409.Gehling, J. G. 1991. The case for Ediacaran fossil roots to the

Metazoan tree. Pp. 181–224 in B. P. Radhakrishna, ed. Theworld of Martin F. Glaessner. Geological Society of IndiaMemoir No. 20. Bangalore.

Gehling, J. G., and J. K. Rigby. 1996. Long expected spongesfrom the Neoproterozoic Ediacara fauna of South Australia.Journal of Paleontology 70:185–195.

Giribet, G., G. D. Edgecombe, and W. C. Wheeler. 2001. Arthro-pod phylogeny based on eight molecular loci and morphol-ogy. Nature 413:157–161.

Glaessner, M. F. 1984. The dawn of animal life: a biohistoricalstudy. Cambridge University Press, Cambridge.

Gould, S. J. 1989. Wonderful life: the Burgess Shale and the na-ture of history. Norton, New York.

———. 1991. The disparity of the Burgess Shale arthropod fau-na and the limits of cladistic analysis: why we must strive toquantify morphospace. Paleobiology 17:411–423.

———. 2002. The structure of evolutionary theory. BelknapPress of Harvard University Press, Cambridge.

Hedges, S. B., and S. Kumar. 2003. Genomic clocks and evolu-tionary timescales. Trends in Genetics 19:200–206.

Hoffman, P. F., A. J. Kaufman, G. P. Halverson, and D. P. Schrag.1998. A Neoproterozoic snowball earth. Science 281:1342–1346.

Holland, N. D., and J.-Y. Chen. 2001. Origin and early evolutionof the vertebrates: new insights from advances in molecularbiology, anatomy, and palaeontology. BioEssays 23:142–151.

Hou, X.-G. 1987. Early Cambrian large bivalved arthropodsfrom Chengjiang, Eastern Yunnan. Acta Paleontological Sin-ica 26:286–297.

Hou, X.-G., and J. Bergstrom. 1997. Arthropods of the Che-ngjiang fauna, southwest China. Fossils and Strata 45:1–116.

Hou, X.-G., L. Ramskold, and J. Bergstrom. 1991. Composition

and preservation of the Chengjiang fauna—a Lower Cambri-an soft-bodied biota. Zoologica Scripta 20:395–411.

Hou, X.-G., J. Bergstrom, and P. Ahlberg. 1995. Anomalocaris andother large animals in the Lower Cambrian Chengjiang Faunaof southwestern China. Geologiska Foreningens Forhandlin-gar 117:163–183.

Hou, X.-G., R. J. Aldridge, D. J. Siveter, D. J. Siveter, and F. Xiang-hong. 2002. New evidence on the anatomy and phylogeny ofthe earliest vertebrates. Proceedings of the Royal Society ofLondon B 269:1865–1869.

Huelsenbeck, J. P., F. Ronquist, R. Neilsen, and J. P. Bollback.2001. Bayesian inference of phylogeny and its impact on evo-lutionary biology. Science 294:2310–4.

Hwang, U. W., M. Friedrich, and D. Tautz. 2001. Mitochondrialprotein joins myriapods with chelicerates. Nature 413:154–7.

Jefferies, R. P. S. 1979. The origin of chordates—a methodolog-ical essay. In M. R. House, ed. The origin of major invertebrategroups. Systematics Association Special Volume 12:443–477.Academic Press, London.

Lacalli, T. C. 2002. Vetulicolians—are they deuterostomes?Chordates? BioEssays 24:208–211.

Lee, M. S. Y. 1992. Cambrian and Recent morphological dispar-ity. Science 258:1816–1817.

Levinton, J. S. 2001. Genetics, paleontology, and macroevolu-tion, 2d ed. Cambridge University Press, Cambridge.

Lofgren, A. S., R. E. Plotnick, and P. J. Wagner. 2003. Morpho-logical diversity of Carboniferous arthropods and insights ondisparity patterns through the Phanerozoic. Paleobiology 29:349–368.

Lynch, M. 1999. The age and relationships of the major animalphyla. Evolution 53:319–25.

Mallatt, J., and J.-Y. Chen. 2003. Fossil sister group of craniates:predicted and found. Journal of Morphology 258:1–31.

Mallatt, J., and J. Sullivan. 1998. 28S and 18S rDNA sequencessupport the monophyly of lampreys and hagfishes. MolecularBiology and Evolution 15:1706–1718.

Mallatt, J., J.-Y. Chen, and N. D. Holland. 2003. Comment on ‘‘Anew species of Yunnanozoan with implications for deutero-stome evolution.’’ Science 300:1372c.

Muller, K. J., D. Walossek, and A. Zakharov. 1995. ‘‘Orsten’’ typephosphatized soft-integument preservation and a new recordfrom the Middle Cambrian Kuonamka Formation in Siberia.Neues Jahrbuch fur Geologie und Palaontologie, Abhandlun-gen 197:101–118.

Patterson, C. 1981. Significance of fossils in determining evo-lutionary relationships. Annual Review of Ecology and Sys-tematics 12:195–223.

Pennisi, E. 2003. Modernizing the tree of life. Science 300:1692–1697.

Peterson, K. J., J. B. Lyons, K. S. Nowak, C, M. Takacs, M. J. War-go, and M. A. McPeek. 2004. Estimating metazoan divergencetimes with a molecular clock. Proceedings of the NationalAcademy of Sciences USA 101:6536–6541.

Peterson, K. J., M. A. McPeek, and D. A. D. Evans. 2005. Tempoand mode of early animal evolution: inferences from rocks,Hox, and molecular clocks. [This volume.]

Ramskold, L. 1992. The second leg row of Hallucigenia discov-ered. Lethaia 25:221–224.

Ramskold, L., and J.-Y. Chen. 1998. Cambrian lobopodians:morphology and phylogeny. Pp. 107–150 in Edgecombe 1998.

Ramskold, L., and X.-G. Hou. 1991. New early Cambrian animaland onychophoran affinities of enigmatic metazoans. Nature351:225–227.

Ronshaugen, M., N. McGinnis, and W. McGinnis. 2002. Hox pro-tein mutation and macroevolution of the insect body plan. Na-ture 415:914–917.

Ruiz-Trillo, I., J. Paps, M. Loukota, C. Ribera, U. Jondelius, J. Ba-guna, and M. Riutort. 2002. A phylogenetic analysis of myo-


sin heavy chain type II sequences corroborates that Acoelaand Nemertodermatida are basal bilaterians. Proceedings ofthe National Academy of Sciences USA 99:11246–11251.

Sanderson, M. J. 2002. Estimating absolute rates of molecularevolution and divergence times: a penalized likelihood ap-proach. Molecular Biology and Evolution 19:101–109.

Seilacher, A. 1989. Vendozoa—organismic construction in theProterozoic biosphere. Lethaia 22:229–239.

Seilacher, A., D. Grazhdankin, and A. Legouta. 2003. Ediacaranbiota: the dawn of animal life in the shadow of giant protists.Paleontological Research 7:43–54.

Selden, P. A., and J. A. Dunlop. 1998. Fossil taxa and relation-ships of chelicerates. Pp. 303–331 in Edgecombe 1998.

Shoemaker, J. S., I. S. Painter, and B. S. Weir. 1999. Bayesian sta-tistics in genetics. a guide for the uninitiated. Trends in Ge-netics 15:354–8.

Shu, D.-G. 2003. A paleontological perspective of vertebrate or-igin. Chinese Science Bulletin 48:725–735.

Shu, D.-G., and S. Conway Morris. 2003. Response to commenton ‘‘A new species of Yunnanozoan with implications for deu-terostome evolution.’’ Science 300:1372d.

Shu, D.-G., S. Conway Morris, and X.-L. Zhang. 1996a. A Pikaia-like chordate from the Lower Cambrian of China. Nature 384:157–158.

Shu, D.-G., X. Zhang, and L. Chen. 1996b. Reinterpretation ofYunnanozoon as the earliest known hemichordate. Nature 380:428–430.

Shu, D.-G., H.-L. Luo, S. Conway Morris, X.-L. Zhang, S.-X. Hu,L. Chen, J. Han, M. Zhu, Y. Li, and L.-Z. Chen. 1999. LowerCambrian vertebrates from south China. Nature 402:42–46.

Shu, D.-G., L. Chen, J. Han, and X.-L. Zhang. 2001a. An EarlyCambrian tunicate from China. Nature 411:472–473.

Shu, D.-G., S. Conway Morris, J. Han, L. Chen, X.-L. Zhang, Z.-F. Zhang, H.-Q. Liu, Y. Li, and J.-N. Liu. 2001b. Primitive deu-terostomes from the Chengjiang Lagerstatte (Lower Cambri-an, China). Nature 414:419–424.

Shu, D.-G., S. Conway Morris, J. Han, Z.-F. Zhang, K. Yasui, P.Janvier, L. Chen, X.-L. Zhang, J.-N. Liu, Y. Li, and H.-Q. Liu.2003a. Head and backbone of the Early Cambrian vertebrateHaikouichthys. Nature 421:526–529.

Shu, D.-G., S. Conway Morris, Z.-F. Zhang, J.-N. Liu, L. Chen,X.-L. Zhang, K. Yasui, and Y. Li. 2003b. A new species of yun-nanozoan with implications for deuterostome evolution. Sci-ence 299:1380–1384.

Simonetta, A. M., and E. Insom. 1993. New animals from theBurgess Shale (Middle Cambrian) and their possible signifi-cance for the understanding of the Bilateria. Bollettino Zool-ogica 60:97–107.

Siveter, D. J., M. Williams, and D. Waloszek. 2001. A phospha-

tocopid crustacean with appendages from the Lower Cam-brian. Science 293:479–80.

Smith, A. B., and K. J. Peterson. 2002. Dating the time of originof major clades: molecular clocks and the fossil record. An-nual Review of Earth and Planetary Sciences 30:65–88.

Smith, M. P., I. J. Sansom, and K. D. Cochrane. 2001. The Cam-brian origin of vertebrates. Pp. 67–84 in Ahlberg 2001.

Sutton, M. D., D. E. G. Briggs, D. J. Siveter, and D. J. Siveter. 2001.An exceptionally preserved vermiform mollusc from the Si-lurian of England. Nature 410:461–3.

Takezaki, N., F. Figueroa, Z. Zaleska-Rutczynska, and J. Klein.2003. Molecular phylogeny of early vertebrates: monophyly ofthe agnathans as revealed by sequences of 35 genes. Molec-ular Biology and Evolution 20:287–292.

Walossek, D. 1999. On the Cambrian diversity of Crustacea. Pp.3–27 in F. R. Schram and J. C. von Vaupel Klein, eds. Proceed-ings of the Fourth International Crustacean Congress, Am-sterdam, The Netherlands, Vol. 1. Brill, Leiden.

Waloszek, D., and J. A. Dunlop. 2002. A larval sea spider (Ar-thropoda: Pycnogonida) from the Upper Cambrian ‘‘Orsten’’of Sweden, and the phylogenetic position of pycnogonids. Pa-laeontology 45:421–446.

Wang, D. Y-C., S. Kumar, and S. B. Hedges. 1999. Divergencetime estimates for the early history of animal phyla and theorigin of plants, animals and fungi. Proceedings of the RoyalSociety of London B 266:163–71.

Wheeler, W. C. 1998. Molecular systematics and arthropods. Pp.9–32 in Edgecombe 1998.

Whittington, H. B., and D. E. G. Briggs. 1985. The largest Cam-brian animal, Anomalocaris, Burgess Shale, British Columbia.Philosophical Transactions of the Royal Society of London B309:569–609.

Wills, M. A. 1998. Cambrian and recent disparity: the picturefrom priapulids. Paleobiology 24:177–199.

Wills, M. A., and R. A. Fortey. 2000. The shape of life: how muchis written in stone? BioEssays 22:1142–1152.

Wills, M. A., D. E. G. Briggs, and R. A. Fortey. 1994. Disparityas an evolutionary index: a comparison of Cambrian and Re-cent arthropods. Paleobiology 20:93–130.

———. 1998. An arthropod phylogeny based on fossil and Re-cent taxa. Pp. 33–106 in Edgecombe 1998.

Winnepenninckx, B. M. H., T. Backeljau, and R. M. Kristensen.1998. Relations of the new phylum Cycliophora. Nature 393:636–637.

Wray, G. A., J. S. Levinton, and L. H. Shapiro. 1996. Molecularevidence for deep Precambrian divergences among metazoanphyla. Science 274:568–72.

Xiao, S., and A. H. Knoll. 2000. Phosphatized animal embryosfrom the Neoproterozoic Doushantuo Formation at Weng’an,Guizhou, South China. Journal of Paleontology 74:767–788.

wonderful strife: systematics, stem groups, and the...

Documents